Hi Ruth,
sorry to insist, but using permutation testing for a single subject vs a group is not valid. Please see my previous email and the archives...
Cheers,
Gwen
--- En date de : Ven 29.8.08, Ruth Carper <[log in to unmask]> a écrit :
> De: Ruth Carper <[log in to unmask]>
> Objet: Re: [FSL] TBSS - single subject vs control group
> À: [log in to unmask]
> Date: Vendredi 29 Août 2008, 21h37
> Did you do both contrasts a minus b and b minus a? The
> stats on tbss, at
> least the old version, were one sided so you just have to
> include both
> contrasts in your matrix.
>
> I.E. 1. -1 on one line and -1 1 on the next line
>
> Dg
> --
>
> Yes, that's what I have. The design files are provided
> below for reference.
> TBSS produced results for the first comparison, but then
> failed with the
> error message:
> terminate called after throwing an instance of
> 'NEWMAT::IndexException'
> Aborted
>
> I suppose I could run it twice, once with the single
> subject first and once
> with him last, but hopefully there's a more efficient
> way.
>
> Command line:
> randomise -i FA_skeletonised.nii -o tbss -m
> mean_FA_skeleton_mask.nii.gz -d
> design.mat -t design.con -n 500 -T 2 -V
>
> The design.con file was:
> /NumWaves 2
> /NumContrasts 2
> /PPheights 1 1
> /Matrix
> 1 -1
> -1 1
>
> And the design.mat was:
> /NumWaves 2
> /NumPoints 8
> /PPheights 1 1
> /Matrix
> 1 0
> 1 0
> 1 0
> 1 0
> 1 0
> 1 0
> 1 0
> 0 1
>
> --Ruth
_____________________________________________________________________________
Envoyez avec Yahoo! Mail. Une boite mail plus intelligente http://mail.yahoo.fr
|